Unverified Commit ecbfe58b authored by Zhiqiang Xie's avatar Zhiqiang Xie Committed by GitHub
Browse files

Bug fix for metrics counter (#4660)

parent 8f163b16
......@@ -878,6 +878,8 @@ class ScheduleBatch(ScheduleBatchDisaggregationDecodeMixin):
# If req.input_embeds is already a list, append its content directly
input_embeds.extend(req.input_embeds) # Use extend to avoid nesting
if req.is_retracted:
req.already_computed = 0
req.cached_tokens += pre_len - req.already_computed
req.already_computed = seq_len
req.is_retracted = False
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment